This study is designed to identity some of the underlying mechanisms that contribute to the etiology of behavior problems. The purpose of this study is threefold: 1) To estimate the incidence and prevalence of behavior problems in the institutionalized long-term care population; 2) To identify the resident and facility correlates and predictors of behavior problems in this population; and 3) To explore the extent to which behavior problems result in hospitalizations and "dumping" (transfer) to other long-term care facilities. Because the prevalence of behavior problems is greater among those in institutions than among those living in the community, this study will focus on residents in nursing homes and personal care homes. Data from the Institutional Component of the 1987 National Medical Expenditures Survey, a nationally representative longitudinal survey of current residents and new admissions to nursing and personal care homes, will be used. These data will permit the generation of estimates of behavior problems for the US nursing home population by subjects, thus providing preliminary information on the characteristics of persons who may be at greater risk of behavioral problems. Data from current residents will be the basis of prevalence estimates and information on new admissions will provide the basis for incidence (new cases) estimates (i.e., persons without behavior problems at admission who later develop behavior problems). Subsequent analyses will include multivariate regression analyses to identify the risk factors of behavior problems. Potential risk factors to be included in these models include resident characteristics of age, sex, race, functional status, incontinence, medical diagnoses/conditions, psychotropic medications, and socialization patterns. Risk factors related to the facility (environmental risk factors) will include facility size, ownership, capacity, type and scope of services provided, resident turnover, percent of private pay residents, presence/type of Alzheimer's units, per diem rate, and staff-patient ratios. Variables derived from these data items at baseline will be used in models predicting behavior problems at follow-up (i.e., 3-9 months later). To assess the impact of behavior problems on risk of hospitalization and "dumping", regression equations will also be estimated where behavior problems become the independent variable and the dependent variable will be utilization or transfer to another facility. Using multiple regression techniques, the independent effect of each risk factor can be ascertained, controlling for other risk factors. The NMES data has only recently been made available and provides a unique opportunity to explore the risk factors associated with behavior problems in a nationally representative sample with longitudinal data using multivariate techniques.